Essentially, known noise reduction algorithms can be traced back to two different and in each case separately used principles, namely diffusion filters on one hand and morphological or edge-preserving filters on the other hand.
Diffusion filters look at the difference between pixel values in relation to statistical uncertainty. As well as the simple diffusion filters, bilateral filters or regularizers used for iterative reconstructions in accordance with patent application DE 10 2009 039 987.9-35 also belong to this class.